Exploring the Efficacy of Binary Surveys versus Likert Scales in Assessing Student Perspectives Using Bayesian Analysis

Abstract

Likert-scale surveys are the undeniable protagonists of online evaluations. They ask the respondent to express their degree of agreement with a series of statements related to the development of a subject. In contrast, in social networks, dichotomous surveys are mostly used. They force respondents to polarize their opinions by selecting “like” or “dislike”. This study compares the efficacy of binary and Likert surveys in gathering student opinions on mechanical engineering program subjects. Using Bayesian analysis, it analyzes the similarity of responses obtained from both formats. For each question and scale, the ratio of “I like” among the total responses collected was calculated. The Bayesian factor method was used to compare the ratios obtained. The null hypothesis was equality between the ratios obtained by the different scales for the same question. This hypothesis was rejected in only 7 of the 49 questions evaluated—less than 15%. Finally, the students were surveyed on the preference for use of both scales. More than 80% stated a preference for the use of the dichotomous format. In view of the results obtained, we recommend more frequent use of the dichotomous scale to gather students’ opinions.